• Proceedings of the Korean Information Science Society Conference
• /
• 2004.04b
• /
• pp.568-570
• /
• 2004

The error-backpropagation (EBP) algerithm for training multilayer perceptrons (MLPs) is known to have good features of robustness and economical efficiency. However, the algorithm has difficulty in selecting an optimal constant learning rate and thus results in non-optimal learning speed and inflexible operation for working data. This paper Introduces an elastic learning rate that guarantees convergence of learning and its local realization by online upoate of MLP parameters Into the original EBP algorithm in order to complement the non-optimality. The results of experiments on a speaker verification system with Korean speech database are presented and discussed to demonstrate the performance improvement of the proposed method in terms of learning speed and flexibility fer working data of the original EBP algorithm.

• Journal of the Korea Society for Simulation
• /
• v.23 no.4
• /
• pp.97-109
• /
• 2014

Personal rapid transit (PRT) is a new transportation system, which is energy efficient and brings high quality of customer service. Customers arrive dynamically at stations and request transportation service. In this paper, we propose a new online PRT dispatching algorithm for pickup and delivery of customers. We adopt the nearest neighbor dispatching rule, which is known as performing well in general. We extend the rule with bipartite matching in order to deal with multiple vehicles and customers at the same time. We suggest a systematic way for selecting vehicles that will be considered to be dispatched, since the scope with which vehicles are selected may affect the system performance. We regard the empty travel distance of vehicles and the customer waiting time as the performance measures. By using simulation experiments, it has been examined that the scope of dispatching affects the system performance. The proposed algorithm has been validated by comparing with other dispatching rules for transportation services. We have shown that our algorithm is more suitable for PRT operating environment than other dispatching rules.

• Smart Structures and Systems
• /
• v.14 no.6
• /
• pp.1247-1267
• /
• 2014

Real-Time Hybrid Simulation (RTHS) is a novel approach conceived to evaluate dynamic responses of structures with parts of a structure physically tested and the remainder parts numerically modelled. In RTHS, delay estimation is often a precondition of compensation; nonetheless, system delay may vary during testing. Consequently, it is sometimes necessary to measure delay online. Along these lines, this paper proposes an online delay estimation method using least-squares algorithm based on a simplified physical system model, i.e., a pure delay multiplied by a gain reflecting amplitude errors of physical system control. Advantages and disadvantages of different delay estimation methods based on this simplified model are firstly discussed. Subsequently, it introduces the least-squares algorithm in order to render the estimator based on Taylor series more practical yet effective. As a result, relevant parameter choice results to be quite easy. Finally in order to verify performance of the proposed method, numerical simulations and RTHS with a buckling-restrained brace specimen are carried out. Relevant results show that the proposed technique is endowed with good convergence speed and accuracy, even when measurement noises and amplitude errors of actuator control are present.

• Journal of KIISE:Computer Systems and Theory
• /
• v.33 no.12
• /
• pp.885-893
• /
• 2006

The Bayesian network(BN) has emerged in recent years as a powerful technique for handling uncertainty iii complex domains. Parameter learning of BN to find the most proper network from given data set has been investigated to decrease the time and effort for designing BN. Off-line learning needs much time and effort to gather the enough data and since there are uncertainties in real world, it is hard to get the complete data. In this paper, we propose an online learning method of Bayesian network parameters from incomplete data. It provides higher flexibility through learning from incomplete data and higher adaptability on environments through online learning. The results of comparison with Voting EM algorithm proposed by Cohen at el. confirm that the proposed method has the same performance in complete data set and higher performance in incomplete data set, comparing with Voting EM algorithm.

• Journal of Digital Convergence
• /
• v.20 no.1
• /
• pp.33-39
• /
• 2022

This study aims to investigate the online class preference depending on students' gender and school level. To achieve this aim, the study conducted a survey on 4,803 elementary, middle, and high school students in 17 regions nationwide. The valid data of 4,524 were then analyzed using the Apriori algorithm to discern the associated patterns of the online class preference corresponding to their gender and school level. As a result, a total of 16 rules, including 7 from elementary school students, 4 from middle school students, and 5 from high school students were derived. To be specific, elementary school male students preferred software-based classes whereas elementary female students preferred maker-based classes. In the case of middle school, both male and female students preferred virtual experience-based classes. On the other hand, high school students had a higher preference for subject-specific lecture-based classes. The study findings can serve as empirical evidence for explaining the needs of online classes perceived by K-12 students. In addition, this study can be used as basic research to present and suggest areas of improvement for diversifying online classes. Future studies can further conduct in-depth analysis on the development of various online class activities and models, the design of online class platforms, and the female students' career motivation in the field of science and technology.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1999.04a
• /
• pp.181-181
• /
• 1999

• Journal of KIISE:Computer Systems and Theory
• /
• v.34 no.10
• /
• pp.493-501
• /
• 2007

The imprecise real-time system provides flexibility in scheduling time-critical tasks. Most scheduling problems of satisfying both 0/1 constraint and timing constraints, while the total error is minimized, are NP-complete when the optional tasks have arbitrary processing times. Liu suggested a reasonable strategy of scheduling tasks with the 0/1 constraint on uniprocessors for minimizing the total error. Song et at suggested a reasonable strategy of scheduling tasks with the 0/1 constraint on multiprocessors for minimizing the total error. But, these algorithms are all off-line algorithms. In the online scheduling, the NORA algorithm can find a schedule with the minimum total error for the imprecise online task system. In NORA algorithm, EDF strategy is adopted in the optional scheduling. On the other hand, for the task system with 0/1 constraint, EDF_Scheduling may not be optimal in the sense that the total error is minimized. Furthermore, when the optional tasks are scheduled in the ascending order of their required processing times, NORA algorithm which EDF strategy is adopted may not produce minimum total error. Therefore, in this paper, an online algorithm is proposed to minimize total error for the imprecise task system with 0/1 constraint. Then, to compare the performance between the proposed algorithm and NORA algorithm, a series of experiments are performed. As a conseqence of the performance comparison between two algorithms, it has been concluded that the proposed algorithm can produce similar total error to NORA algorithm when the optional tasks are scheduled in the random order of their required processing times but, the proposed algorithm can produce less total error than NORA algorithm especially when the optional tasks are scheduled in the ascending order of their required processing times.

• Journal of Korean Institute of Industrial Engineers
• /
• v.35 no.3
• /
• pp.213-217
• /
• 2009

We consider the online scheduling problems of three and four machines under eligibility constraint. Respectively for the cases of three and four machines, we prove that AW algorithm has competitive ratios of $\frac{5}{2}$ and 3 which are shown to be optimal. Also, we show that the same results hold for the semi-online cases with prior knowledge of the total and the largest processing time.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2004.10a
• /
• pp.471-474
• /
• 2004

In this paper, a NURBS surface interpolator is proposed which can deal with shapes defined from CAD/CAM programs on a surface basis and can improve contour accuracy. The proposed interpolator is based on newly defined G-codes and includes online tool-path planning suitable for NURBS surface machining. The real-time interpolation algorithm, considering an effective machining method for each machining process and minimum machining time, is executed in an online manner. The proposed interpolator is implemented on a PC-based 3-axis CNC milling system and evaluated through actual machining in terms of machining time and regulation of feedrate and cutting force in comparison with the existing method.

• ETRI Journal
• /
• v.34 no.2
• /
• pp.288-291
• /
• 2012

In this letter, a new online anomaly detection approach for software systems is proposed. The novelty of the proposed approach is to apply a new semantic kernel function for a support vector machine (SVM) classifier to detect fault-suspicious execution paths at runtime in a reasonable amount of time. The kernel uses a new sequence matching algorithm to measure similarities among program execution paths in a customized feature space whose dimensions represent the largest common subpaths among the execution paths. To increase the precision of the SVM classifier, each common subpath is given weights according to its ability to discern executions as correct or anomalous. Experiment results show that compared with the known kernels, the proposed SVM kernel will improve the time overhead of online anomaly detection by up to 170%, while improving the precision of anomaly alerts by up to 140%.